This study examined the extent to which instructional conditions influence the prediction of academic success in nine undergraduate courses offered in a blended learning model (n=4134).
The results of the study indicate that using generalized models to predict academic success prediction threatens the potential of learning analytics to improve the quality of teaching and learning practice. This is consistent with literature on learning theory, which stresses the importance of contextual factors when considering academic risk.
The authors suggest that the under-explored role of contextual variables may help explain the mixed findings in the learning analytics field. To date, even large-scale studies have reported differences in the overall predictive power of the same variables associated with activity in learning management systems (LMS).
The authors conclude that it is crucial for learning analytics research to take into account the diverse ways in which technology is adopted and applied in course-specific contexts. A lack of attention to instructional conditions can result in over- or under-estimation of the effects of LMS features on academic success.
The authors suggest the use of learning analytics studies designed with clear theoretical frameworks that will a) connect learning analytics research with decades of previous research in education and b) make clear what is contended by research designs, and so make explicit what the research outcomes mean in relation to existing models and previous findings.
In relation to the propositions of this Evidence Hub, the paper notes that, “Learning analytics will not be of practical value or widely adopted if it cannot offer insights that are useful for both learners and teachers”.